This dataset is based on Landsat remote sensing images. Through preprocessing algorithms such as radiometric calibration and atmospheric correction, images of desert areas were obtained. The boundaries of the desert areas were extracted through manual visual interpretation and band index methods. The deserts involved in this dataset mainly include the major deserts along the "Belt and Road" region. They are the Tengger Desert, the Taklimakan Desert, the Badain Jaran Desert, the Kubuqi Desert, the Ulan Buh Desert, the Kumtag Desert, the Gurbantunggut Desert, the Qaidam Desert,Atyrau desert,Karakum desert,Kavir desert,Kharan desert,Kyzylkum desert,Lut desert,Moinkum desert,Registan desert,Thar desertand a total of 17 deserts. The time span is from 2000 to 2020, with data collected every five years.
1. Name of Data
The regions along the Belt and Road Initiative 2000-2020 Boundary Dataset
2. Data description of attribute items
Data accuracy: 30m
Landsat Images
Download from the websites of the United States Geological Survey (https://www.usgs.gov/) and Geospatial Data Cloud (http://www.gscloud.cn/).
1. Remote sensing data preprocessing: Radiation calibration, atmospheric correction, image stitching, cropping, color correction.
2. Manual annotation method for extracting desert boundary: Using labelme based annotation software tools to manually annotate desert boundaries on cropped images as the training set for the model.
3. Data augmentation: Perform mirror flipping and directional rotation on the training set to obtain a dataset that is 10 times larger.
The average accuracy of the marked desert boundary is within 1 pixel (30 m).
| # | number | name | type |
| 1 | 2022YFF0711702-04 | National key R & D plan |
This work is licensed under a
Creative
Commons Attribution 4.0 International License.
CSTR:11738.11.NCDC.NIEER.DB6904.2025
10.12072/ncdc.nieer.db6904.2025